| 1 |
Haveged, an entropy source |
| 2 |
|
| 3 |
INTRODUCTION |
| 4 |
|
| 5 |
Complete documentation on haveged can be found at http://www.issihosts.com/haveged/ |
| 6 |
|
| 7 |
Linux provides device interfaces (/dev/random and /dev/urandom) to a pool of |
| 8 |
random numbers collected from system interrupt service routines. On some |
| 9 |
systems, especially on those systems with high needs or limited user |
| 10 |
interaction, the standard collection mechanism cannot meet demand. In those |
| 11 |
cases, an adequate supply of random numbers can be maintained by feeding |
| 12 |
additional entropy into /dev/random pool via a file system interface. The |
| 13 |
haveged daemon was created to fulfill this function using random data generated |
| 14 |
by the HAVEGE algorithm. |
| 15 |
|
| 16 |
The HAVEGE algorithm is based upon the indirect effects of unrelated hardware |
| 17 |
events on the instruction timing of a calculation that is sensitive to processor |
| 18 |
features such as branch predictors and instruction/data access mechanisms. |
| 19 |
Samples from a high-resolution timer are input into the algorithm to produce a |
| 20 |
stream of random data in a collection buffer. The algorithm requires no special |
| 21 |
privilege other than access to a high resolution timer, as provided by hardware |
| 22 |
instruction or a system call. |
| 23 |
|
| 24 |
The HAVEGE mechanism is implemented in C using in-line assembly only where |
| 25 |
direct hardware access is needed. On modern compilers, compiler intrinsics are |
| 26 |
used to replace much if not all in-line assembly. The haveged implementation of |
| 27 |
HAVEGE adds two crucial features: automated tuning of the algorithm to an |
| 28 |
environment and a run time facility to verify the data generated in the |
| 29 |
collection buffer. |
| 30 |
|
| 31 |
The haveged collection loop is tuned at run-time to match the size of the hosts |
| 32 |
L1 data and instruction caches. The size determination is made on the basis of |
| 33 |
the best fit to the following (low to high): |
| 34 |
|
| 35 |
1. as a compiled default |
| 36 |
2. as determined by cpuid, if available |
| 37 |
3. as determined by the /sys file system, if available |
| 38 |
4. as specified by initialization parameters. |
| 39 |
|
| 40 |
Run-time verification of collection buffer contents is based upon the |
| 41 |
methodology from the proposed AIS-31 standard from BSI, the German Federal |
| 42 |
Office for Information Security (Bundesamt für Sicherheit in der |
| 43 |
Informationstechnik). In the suggested configuration, haveged passes most of the |
| 44 |
requirements for a NTG.1 class device described in version two of the AIS-31 |
| 45 |
specification. |
| 46 |
|
| 47 |
The haveged implementation also provides an experimental feature to multiplex |
| 48 |
havege collections over multiple cpu cores. Interprocess coordination in this |
| 49 |
configuration sacrifices a significant fraction of haveged throughput in |
| 50 |
exchange for a cpu load spread over a number of processors. |
| 51 |
|
| 52 |
All of the above features are contained within the haveged-devel sub package of |
| 53 |
haveged. The haveged package can be built with libtool to expose the devel sub |
| 54 |
package or without libtool if the devel library is not needed. In either case, |
| 55 |
the executable built contains a file system interface to haveged-devel features |
| 56 |
and an optional facility to feed havege output into the random device. The build |
| 57 |
system also provides "check" targets to test the output of the executable's |
| 58 |
random number generator through the file system interface. |
| 59 |
|
| 60 |
TESTING haveged |
| 61 |
|
| 62 |
Testing via build check targets has been part of the haveged distribution since |
| 63 |
the project moved to automake. Original tests were limited to a quick evaluation |
| 64 |
implemented as an adaptation of the open source ent tool from Fourmilab and a |
| 65 |
more thorough evaluation based upon the NIST SP800 test suite. Run time testing |
| 66 |
was added to haveged version 1.5 as part of an effort to loosely follow the RNG |
| 67 |
testing AIS-31 framework of the German Common Criteria organization, BSI. |
| 68 |
|
| 69 |
The significant features of the AIS-31 framework are: |
| 70 |
|
| 71 |
Run-time tests are broken into 2 groups, Procedure A containing 5 tests, and |
| 72 |
Procedure B containing 3 tests. Procedure A consists of performing a 48-bit |
| 73 |
disjointedness test on 64K sequences, followed by 257 repetitions of the four |
| 74 |
FIPS-140-1 tests and an auto-correlation test on successive 2000 bit sequences. |
| 75 |
Procedure B performs distribution tests for 10,000 occurrences of 1, 2, 3, 4 bit |
| 76 |
runs in successive samples, followed by a entropy estimate based upon on another |
| 77 |
256000+2560 bit sample. A sample must pass all individual tests to pass the |
| 78 |
procedure. An ideal RNG is expected to pass Procedure A with a probability of |
| 79 |
0.9987 and pass Procedure B with with a probability of 0.9998. One retry will be |
| 80 |
attempted to recover from the the failure of a single test in either procedure. |
| 81 |
The probability an ideal generator would fail the retry is nill. |
| 82 |
|
| 83 |
Special features of the haveged implementation: |
| 84 |
|
| 85 |
1) A failure of either procedure is a fatal error. In the case of haveged, the |
| 86 |
instance will terminate. |
| 87 |
2) The tests operate directly on the collection buffer after a fill. Performance |
| 88 |
costs are acceptable except for the auto-correlation test, test5. An option |
| 89 |
to skip test5 on some repetitions mitigates this problem. |
| 90 |
3) Tests can take place at start up (a "tot" test) or continuously as specified |
| 91 |
by a haveged parameter. In the continuous case, there is no implied alignment |
| 92 |
between the collection buffer and the testing context. Collection buffer |
| 93 |
contents will be released for consumption provided the buffer does not contain |
| 94 |
a failed individual test. |
| 95 |
4) The size of the input required to complete procedure B depends on the content. |
| 96 |
This means there is no fixed alignment of the test input in the collection |
| 97 |
buffer. |
| 98 |
5) Procedure retries are logged. Extended information is available with -v3. Retries |
| 99 |
are expected (see failure rates above) but normally only seen with output |
| 100 |
ranges north of a few GB. |
| 101 |
|
| 102 |
More detailed information on the adaptation of the BIS framework can be found |
| 103 |
at http://www.issihosts.com/haveged/ais31.html |
| 104 |
|
| 105 |
BUILDING haveged |
| 106 |
|
| 107 |
This package originated on "Enterprise Linux 5" systems (RHEL 5 / CentOS 5 / SL |
| 108 |
5), but every effort has been made to retain and broaden the hardware support of |
| 109 |
the original HAVEGE implementation. The package uses the automake build system. |
| 110 |
By default, the build uses a libtool build to expose haveged-devel. The |
| 111 |
contrib/build directory contains a build.sh script to toggle the libtool |
| 112 |
requirement and recover from some autotools failures. Change directory to the |
| 113 |
build directory and type ./build.sh for options. |
| 114 |
|
| 115 |
The configure process uses hardware detection via config.sub or the configure |
| 116 |
"-host" command line argument. The configure "host" variable is used to select |
| 117 |
in-line assembly or compiler intrinsics appropriate to the build target. |
| 118 |
|
| 119 |
Currently supported hosts are: |
| 120 |
|
| 121 |
1. x86 (x86_64 and i686) |
| 122 |
2. aarch64 |
| 123 |
3. armv7hl |
| 124 |
4. ppc64le |
| 125 |
5. ia64 |
| 126 |
6. powerpc |
| 127 |
7. s390 |
| 128 |
|
| 129 |
|
| 130 |
The generic host type is provided for those systems without user level access to |
| 131 |
a high-resolution system timer. In this case, the --enable-clock_gettime option |
| 132 |
is set to 'yes'. |
| 133 |
|
| 134 |
The following build options are available to "./configure": |
| 135 |
|
| 136 |
1. --enable-clock_gettime (default 'no' for recognized hosts) |
| 137 |
2. --enable-daemon (default 'yes' if Linux) |
| 138 |
3. --enable-diagnostic (default 'no') |
| 139 |
4. --enable-nistest (default 'no' but recommended) |
| 140 |
5. --enable-olt (default 'yes') |
| 141 |
6. --enable-threads (experimental) |
| 142 |
7. --enable-tune (default 'yes') |
| 143 |
|
| 144 |
Detailed option information is available by typing "./configure --help". For |
| 145 |
options xxx that take "yes/no" arguments, --disable-xxx may be used as the |
| 146 |
inverse of --enable-xxx. |
| 147 |
|
| 148 |
If --enable-clock_gettime() is 'yes', the clock_gettime(CLOCK_MONOTONIC) system |
| 149 |
call will be used as a timer source. This option defaults to 'yes' for generic |
| 150 |
host builds and 'no' otherwise. This option may proved useful if access to time |
| 151 |
hardware is privileged. Due to variability of clock_gettime() implementations, |
| 152 |
the adequacy of the clock_gettime() resolution cannot be known until run time. |
| 153 |
It is strongly advised to use run-time testing if this option is used. |
| 154 |
|
| 155 |
If --enable-daemon is 'yes', ioctl access required to the random device and |
| 156 |
read-write access to the /proc virtual file system is required. The daemon may |
| 157 |
be run in the foreground or fork into the background based upon a command line |
| 158 |
argument. The daemon interface targets the 2.6 and later kernel and may not work |
| 159 |
on 2.4 kernels due to difference in the random interface between those two |
| 160 |
kernel versions. The change in the proc file system from pool size expressed in |
| 161 |
bytes to pool size expressed in bits has been taken into account, other changes |
| 162 |
may be required. This option is 'no' when diagnostic modes are enabled. If |
| 163 |
the option is no the executable is installed in the user bin directory instead |
| 164 |
of the sbin directory. |
| 165 |
|
| 166 |
If --enable-diagnostic is 'yes', the capture and inject diagnostic interfaces |
| 167 |
are enabled. The capture or inject diagnostic may be enabled singly by setting |
| 168 |
the option to 'capture' or 'inject'. A setting for any value other than 'no' |
| 169 |
for this option forces --enable-daemon=no. See DIAGNOSTICS below for details. |
| 170 |
|
| 171 |
The --enable-nistest option enables more thorough testing for the check target. |
| 172 |
See CHECKING for details. |
| 173 |
|
| 174 |
The --enable-olt option is provided to suppress the entire online test facility. |
| 175 |
This option is provided for systems with a very limited resource budget and the |
| 176 |
ability to thoroughly test the RNG output by other means. When enabled, the online |
| 177 |
test system tests the output of the haveged random number generator using AIS-31 |
| 178 |
test procedures A and B. Either or both tests may be run as a total failure check |
| 179 |
(a "tot" test) at initialization and/or continuously during all subsequent haveged |
| 180 |
operation - See the man page and the description at |
| 181 |
http://www.issihosts.com/haveged/ais31.html for further information. |
| 182 |
|
| 183 |
The --enable-threads option is an experimental prototype for running multiple |
| 184 |
collection threads in a single haveged instance. The goal is to create a |
| 185 |
multi-core haveged that would spread collection overhead more evenly over the |
| 186 |
available cpu resources. |
| 187 |
|
| 188 |
The --enable-tune option allows the on-line tuning facility to be suppressed. |
| 189 |
This is intended for systems with special needs and or a limited resource budget. |
| 190 |
Setting the option to 'yes' enables both the cpuid and virtual file system methods, |
| 191 |
a value of 'no' suppresses both methods. Individual tuning methods can be selected |
| 192 |
by setting the option to either 'cpuid' or 'vfs'. Note that the 'cpuid' method |
| 193 |
is always conditional on host type and will not be present if the hardware |
| 194 |
architecture does not support the instruction. |
| 195 |
|
| 196 |
The haveged build does not modify CFLAGS and DFLAGS, so for example: |
| 197 |
|
| 198 |
CFLAGS="-fpic -DGENERIC_DCACHE=32" LDFLAGS="-z now" ./configure --disable-tune |
| 199 |
|
| 200 |
would configure the build for a manually-tuned, hardened daemon with a default |
| 201 |
data cache size of 32kb. |
| 202 |
|
| 203 |
|
| 204 |
CHECKING haveged |
| 205 |
|
| 206 |
The build check target provides two test procedures for the build. |
| 207 |
|
| 208 |
1. A "quick" check based upon and adaptation of the public domain ENT program. |
| 209 |
The "entest" program uses the ENT sources to subject a sample to the following: |
| 210 |
|
| 211 |
a) The Chi-Square result must fall within acceptable bounds (>1% and <99 %) |
| 212 |
b) The entropy/character must exceed a minimum (7.5) |
| 213 |
c) The arithmetic mean must exceed a minimum (127.0) |
| 214 |
d) The monte-carlo approximation of PI must lie within error bounds (.5%) |
| 215 |
e) The Sequential Correlation Coefficient must be below a minimum (.8) |
| 216 |
|
| 217 |
The program provides a pass-fail indication and an optional display of the |
| 218 |
results to stdout. The entest checks are quite weak and intended only as a |
| 219 |
quick sanity check. |
| 220 |
|
| 221 |
2. An adaptation of the NIST Statistical Test Suite as adapted by Oliver |
| 222 |
Rochecouste of irisa.fr as part of the original havege project. More that 400 |
| 223 |
tests are performed in a typical run. The program provides as pass-fail |
| 224 |
indication with detailed results reported in the nist.out file in the |
| 225 |
nist directory. You will need sit down with SP800-*.pdf available from the |
| 226 |
NIST to review the detailed results. AIS31 provides recommendations for the |
| 227 |
NIST test suite as 'additional tests'. See testing documentation at |
| 228 |
http://www.issihosts.com/haveged/ais31.html for further information. |
| 229 |
|
| 230 |
The "quick" test is always part of the check target. The NIST suite is run only |
| 231 |
when --enable-nistest is 'yes'. |
| 232 |
|
| 233 |
Both checks function the same way, haveged is run to collect a sample file in |
| 234 |
the test directory which is then analyzed by the test program. A pass-fail return |
| 235 |
is given in both cases, additional information is written to stdout. The input |
| 236 |
samples and the nist.out report are deleted by the clean make target. |
| 237 |
|
| 238 |
Both test mechanisms are statistical and even a fully functional random number |
| 239 |
generator will experience occasional failures. It is not uncommon to see one or |
| 240 |
two failures in the NIST suite and the entest will occasionally fail with a small |
| 241 |
sample size (usually the Chi-Square test barks). Early haveged releases used a |
| 242 |
entest sample size of 1MB, this has been increased to 16MB because failures with |
| 243 |
that sample size were all too common. A 16MB sample will also deplete and refill |
| 244 |
the haveged collection area to exercise all buffer logic. |
| 245 |
|
| 246 |
Users are encouraged to run their own external tests. The --number==0 option is |
| 247 |
a convenient means to pipe haveged output into external suites such as Dieharder, |
| 248 |
the TESTU01 batteries, or PractRand. |
| 249 |
|
| 250 |
|
| 251 |
RUNNING haveged |
| 252 |
|
| 253 |
The following invocation arguments are always available: |
| 254 |
|
| 255 |
--buffer , -b [] Buffer size [KW] - default : 128 |
| 256 |
--data , -d [] Data cache size [KB], with fallback to 16 |
| 257 |
--inst , -i [] Instruction cache size [KB], with fallback to 16 |
| 258 |
--file , -f [] Sample output file - default: 'sample', '-' for stdout |
| 259 |
--number , -n [] Output size in [k|m|g|t]. 0 = unlimited to stdout |
| 260 |
--verbose , -v [] Verbose mask 0=none,1=summary,2=retries,4=timing,8=loop,16=code,32=test |
| 261 |
--help , -h This help |
| 262 |
|
| 263 |
The "-b", "-d", "-i" options are needed only in special cases. Generator output |
| 264 |
should be validated after changes to these values. |
| 265 |
|
| 266 |
If haveged is built with online testing enabled, the following is present |
| 267 |
|
| 268 |
--onlinetest , -o [] [t<x>][c<x>] x=[a[n][w]][b[w]] 't'ot, 'c'ontinuous, default: ta8b" |
| 269 |
|
| 270 |
The default configuration executes the "tot" test using AIS procedure B. At the completion |
| 271 |
of the tot test, the buffer is reloaded and any continuous tests will be run before |
| 272 |
data becomes available. |
| 273 |
|
| 274 |
If haveged is built with threads support, the following is present |
| 275 |
|
| 276 |
--threads , -t [] Number of threads |
| 277 |
|
| 278 |
If daemon interface is enabled, the following options are available: |
| 279 |
|
| 280 |
--Foreground , -F Run daemon in foreground, do not fork and detach, |
| 281 |
--pid , -p [] The location of the daemon pid file, default: /var/run/haveged.pid |
| 282 |
--run , -r [] 0=daemon,1=config info,>1=Write <r>KB sample file |
| 283 |
--write , -w [] Set write_wakeup_threshold [bits] |
| 284 |
|
| 285 |
If haveged is running detached, diagnostic output is written to syslog, otherwise |
| 286 |
output is written to stderr. |
| 287 |
|
| 288 |
If the daemon interface is enabled, non-zero "-r" options are used to test the |
| 289 |
haveged random number generator; the random number generator will be configured, |
| 290 |
the initial data collection pass will be executed, configuration details will be |
| 291 |
written to stdout, and a "-r" KB sample of output will be written to the sample |
| 292 |
output file for all "-r" > 1. The "-n" option provides a more friendly version |
| 293 |
of r > 1. |
| 294 |
|
| 295 |
If the daemon interface and --run==1 is specified, or if --run is not available |
| 296 |
and --number is not specified a summary of build, tuning, and execution |
| 297 |
is displayed and haveged exits without generating data: |
| 298 |
|
| 299 |
<prog>: ver: <ver>; arch: <arch>; vend: <vend>, build: (<opts>); collect: <collect> |
| 300 |
<prog>: cpu: <cpu> (<tune>);data: <data> (<tune>); inst: <inst> (<tune>); idx: <idx>; sz: <sz> |
| 301 |
<prog>: tot tests(<spec>): <score>; continuous tests(<spec>):<score> last entropy estimate <ent> |
| 302 |
<prog>: fills: <fills>, generated: <total> |
| 303 |
|
| 304 |
where |
| 305 |
<prog> build: program name - normally haveged |
| 306 |
<ver> build: package version |
| 307 |
<arch> build: architecture: ia64, generic, ppc, s390, sparc, sparclite, x86 |
| 308 |
<vend> build: vendor string of host from cpuid if available else 'generic' |
| 309 |
<opts> build: version of gcc used and build option flags - see below. |
| 310 |
<collect> tuning: collection buffer size |
| 311 |
<cpu> tuning: (sources list for cpu info, see below) |
| 312 |
<data> tuning: size, (sources list for L1 data cache) |
| 313 |
<inst> tuning: size, (sources list for L1 instruction cache) |
| 314 |
<idx> tuning: collector loops used/collector loops available |
| 315 |
<sz> tuning: collector size used/collector size available. |
| 316 |
<tune> tuning: tuning sources - see below |
| 317 |
<spec> exec: tests to be executed in --onlinetest format |
| 318 |
<ent> exec: last entropy estimate from procedure B. |
| 319 |
<score> exec: pass/fail counts for AIS test procedures |
| 320 |
<fills> exec: number of times buffer was filled |
| 321 |
<total> exec: number of bytes output |
| 322 |
|
| 323 |
build option flags represent the ./configure options as: |
| 324 |
C=clock_gettime, D=diagnostic I=tune with cpuid, M=multi-core, T=online tests, V=tune with vfs |
| 325 |
|
| 326 |
tuning sources are: |
| 327 |
D=default value, P=instance parameter, C=cpuid present, |
| 328 |
H=hyperthreading, A=AMD cpuid, A5=AMD fn5, A6=AMD fn6, A8=AMD fn8 |
| 329 |
L2=Intel has leaf2, L4=Intel has leaf4, B=Intel leaf b, |
| 330 |
4=intel leaf4, V=virtual file system available |
| 331 |
VS=/sys/devices/system/cpu/cpu%d/cache/index<n>/level, |
| 332 |
VO=/sys/devices/system/cpu/online, VI=/proc/cpuinfo |
| 333 |
VC=/sys/devices/system/cpu |
| 334 |
|
| 335 |
Sources displayed in parenthesis are white space separated lists of the above |
| 336 |
tokens. The "collector * used/collector *available" values indicate the fit of |
| 337 |
the haveged collection loop to the L1 instruction cache. |
| 338 |
|
| 339 |
Items marked 'exec:' above have meaning only when data is generated. The display |
| 340 |
of test results is customized to match the test options specified. The last few |
| 341 |
lines of the build/tuning summary (those items marked 'exec:' above) also |
| 342 |
appear by themselves as the 'exec summary' in other circumstances. |
| 343 |
|
| 344 |
In other circumstances, the completion of initialization is announced by a banner |
| 345 |
written to the log: |
| 346 |
|
| 347 |
"<prog>: starting up" |
| 348 |
"Writing <n> byte output to <file>" |
| 349 |
|
| 350 |
The first line above is used when running as a daemon, the second otherwise. If |
| 351 |
one or more online tests are running, test failures are logged as: |
| 352 |
|
| 353 |
<prog>: AIS-31 <stage> procedure <proc>: <action> <bytes> bytes fill <fill> |
| 354 |
|
| 355 |
where |
| 356 |
<stage> is either 'tot' or 'continuous' |
| 357 |
<proc> is either 'A' or 'B' |
| 358 |
<action> is either 'retry' or 'fail' |
| 359 |
<bytes> is number of bytes processed in procedure before failure |
| 360 |
<fill> is the number of times the buffer was filled |
| 361 |
|
| 362 |
The exec summary is logged upon error or signal terminations. Other log output is |
| 363 |
controlled by --verbose: |
| 364 |
|
| 365 |
A --verbose bit mask to obtain additional diagnostic information: |
| 366 |
|
| 367 |
0x01 Show exec summary on termination, retry summary |
| 368 |
0x02 Show online test retry details |
| 369 |
0x04 Show timing for collections |
| 370 |
0x08 Show collection loop characteristics |
| 371 |
0x10 Show code offsets |
| 372 |
0x20 Show all online test completion details |
| 373 |
|
| 374 |
The "--write" option will set proc/sys/kernel/random/write_wakeup_threshold to |
| 375 |
the given value. This is useful because this threshold is very small on some |
| 376 |
systems. A minimum of 1024 is recommended. |
| 377 |
|
| 378 |
The file system interface supports file creation of up data setups up to 16tb or |
| 379 |
can be part of a piped command set. See the man(8) page for examples. |
| 380 |
|
| 381 |
DIAGNOSTIC haveged build |
| 382 |
|
| 383 |
The diagnostic build is a special version of the non-daemon configuration with a |
| 384 |
specialized --run option. The --run levels 0 and 1 correspond duplicate the |
| 385 |
equivalent application functions. One or more of the addition commands may be |
| 386 |
included depending on the --enable-diagnostic value chosen. |
| 387 |
|
| 388 |
capture (--run 2) |
| 389 |
Run the RNG but collect the timer inputs in a separate buffer. The normal |
| 390 |
RNG output is discarded and replaced by the timer inputs. Requires additional |
| 391 |
buffering, performance is poor because 8 times as many fills are required to |
| 392 |
generate same amount of output. |
| 393 |
|
| 394 |
inject tics (--run 4) |
| 395 |
Replace timer source with file input. Requires additional buffering, source |
| 396 |
must account for generator warmup. |
| 397 |
|
| 398 |
inject data (--run 8) |
| 399 |
Bypass the normal RNG completely and read input directly into the output |
| 400 |
buffer. Generator warmup is skipped. This option is useful for validating |
| 401 |
the online test suite. |
| 402 |
|
| 403 |
Knowledge of haveged internals is needed to use the special features effectively |
| 404 |
and interpret the results. The diagnostic build identifies itself as |
| 405 |
'havege_diagnostic' in -help display and other outputs. |
| 406 |
|
| 407 |
|
| 408 |
INSTALLATION |
| 409 |
|
| 410 |
If the daemon interface is not enabled, the install places the executable in |
| 411 |
automake's bin_PROGRAMS directory and provides a man(8) page. A man(3) page |
| 412 |
is provided for the libtool build. If the daemon interface is enabled, the |
| 413 |
executable is installed in automake's sbin_PROGRAMS directory. |
| 414 |
|
| 415 |
EXTRAS |
| 416 |
|
| 417 |
The contrib directory contains bits and pieces that are not integrated into the |
| 418 |
distribution. Currently this directory contains build related utilities in |
| 419 |
the build directory and an unorganized collection of some of the tools used |
| 420 |
to analyze haveged in the diags directory. |
| 421 |
|
| 422 |
The script contrib/build/build.sh organizes the build utilities. The script |
| 423 |
provides methods to toggle the use of libtool in the build in particularly |
| 424 |
problematic environments, a bootstrap to recover from automake error states |
| 425 |
and the option to build and run a devel sample program after the devel |
| 426 |
sub package has been built. |
| 427 |
|
| 428 |
Several sample package spec files are provided as contrib/build/*.spec. The |
| 429 |
havege_sample.c source demonstrates usage of the havege.h API. |
| 430 |
|
| 431 |
|